Kibana and Edge-LB

Exposing Kibana using Edge-LB

Expose Kibana using Edge-LB

Install Edge-LB

  1. First, check if Edge-LB is available on your DC/OS cluster by running:

    dcos package search edgelb

    The output should look something like:

    $ dcos package search edgelb
    edgelb       v1.3.0   True      False      EdgeLB on DC/OS
    edgelb-pool  v1.3.0   True      True       EdgeLB Pool on DC/OS

    If it does, you can skip the dcos package repo add commands below.

    Otherwise, if you see a No packages found message, add a couple of package repositories to your cluster. For information about the current Edge-LB version support and compatibility, see the Edge-LB documentation and the Certified packages and DC/OS versions to compatibility matrix.

  2. Use commands similar to the following to install the Edge-LB packages:

    dcos package repo add edgelb
    dcos package repo add edgelb-pool
  3. Now install Edge-LB with:

    dcos package install edgelb

    For more information about installing and configuring Edge-LB, see the installation instructions in the Edge-LB documentation.

Create an Edge-LB pool for Kibana

The following command will create an Edge-LB pool task running on one of your DC/OS cluster’s public agents, which will allow Kibana to be accessed from outside the cluster network, given that the selected port on the agent machine is open.

In this example, we’ll expose a Kibana service named /production/kibana through HTTP, on port 80. It will be accessible on http://$public_agent_ip_or_url:80.

Note that in this example:

  • the Edge-LB pool that will be created is named kibana
  • its backend name is kibana-backend

It is not a requirement that these match any configuration options related to the actual Kibana service, so you could name them differently.

The pool fields that actually map to the actual Kibana service are under haproxy.backends:

  • rewriteHttp.path.fromPath should match the Kibana Marathon app service path
  • services.endpoint.portName should match the Kibana Marathon app port name
  • services.marathon.serviceID should match the Kibana service name
  1. Let’s get the remaining configuration parameters that will map the Edge-LB pool to the actual Kibana service. We’ll use them in the pool configuration. Make sure to use a different name or port based on your needs.

    kibana_port_name="$(dcos marathon app show "${kibana_service_name}" | jq -r '.portDefinitions[0].name')"
    echo "{
      \"apiVersion\": \"V2\",
      \"role\": \"slave_public\",
      \"name\": \"kibana\",
      \"count\": 1,
      \"haproxy\": {
        \"stats\": {
          \"bindPort\": 9090
        \"frontends\": [
            \"bindPort\": ${kibana_proxy_port},
            \"linkBackend\": {
              \"defaultBackend\": \"kibana-backend\"
            \"protocol\": \"HTTP\"
        \"backends\": [
            \"name\": \"kibana-backend\",
            \"protocol\": \"HTTP\",
            \"rewriteHttp\": {
              \"path\": {
                \"fromPath\": \"${kibana_service_path}\",
                \"toPath\": \"/\"
            \"services\": [
                \"marathon\": {
                  \"serviceID\": \"${kibana_service_name}\"
                \"endpoint\": {
                  \"portName\": \"${kibana_port_name}\"
    }" > kibana_pool.json

    Which will end up looking like:


      "apiVersion": "V2",
      "role": "slave_public",
      "name": "kibana",
      "count": 1,
      "haproxy": {
        "stats": {
          "bindPort": 9090
        "frontends": [
            "bindPort": 80,
            "linkBackend": {
              "defaultBackend": "kibana-backend"
            "protocol": "HTTP"
        "backends": [
            "name": "kibana-backend",
            "protocol": "HTTP",
            "rewriteHttp": {
              "path": {
                "fromPath": "/service//production/kibana",
                "toPath": "/"
            "services": [
                "marathon": {
                  "serviceID": "/production/kibana"
                "endpoint": {
                  "portName": "kibana"
  2. Now you can install the Kibana Edge-LB pool with:

    dcos edgelb create kibana_pool.json

    Again, installation will take a moment. If TASK_RUNNING appears in the output of the following command, it means that the pool is up and running.

    dcos edgelb status kibana

At this point, Kibana should already be accessible through http://$public_agent_ip_or_url:80.

Accessing Kibana

If you only have one public agent and you know its IP address, it should be easy to access Kibana. If not, there are a few commands that might help.

Get IP address of public agent running Kibana pool

This step requires that you have SSH access to the DC/OS cluster nodes. Make sure you do before proceeding.

Here we’re using the kibana pool name in the dcos edgelb status command. If you named the pool something else, make sure to use it instead.

agent_private_ip="$(dcos edgelb status kibana --json | jq -r '.[0].status.containerStatus.networkInfos[0].ipAddresses[0].ipAddress')"
agent_public_ip="$(dcos node ssh --option StrictHostKeyChecking=no --option LogLevel=quiet --master-proxy --private-ip="${agent_private_ip}" "curl -s")"

Authenticate with Kibana

Now that we have the public agent IP address where the Edge-LB Kibana pool task is running, we should be able to access Kibana.

  1. If Kibana has X-Pack Security enabled, you’ll first need to access http://$public_agent_ip_or_address/login to authenticate with the Kibana server. Use credentials that are stored in your Elasticsearch cluster.

    command -v xdg-open && xdg-open "${kibana_login_url}" || open "${kibana_login_url}"
  2. After authenticating, or if Kibana doesn’t have X-Pack Security enabled, Kibana should be available at http://$public_agent_ip_or_url/service/kibana/app/kibana.

    command -v xdg-open && xdg-open "${kibana_authenticated_url}" || open "${kibana_authenticated_url}"